The AI wants to rewrite a large section to fix a small bug. What's the risk?

You discard working code and add new risk in exchange for one fix

Why does 'that didn't work' make debugging worse?

It carries no information — no error, no repro, nothing to distinguish theories, so the AI guesses again

Vibecoding · Module 07 of 7

Maintenance

Keeping a vibe coded app working — debugging with the AI, and stopping small changes from breaking everything.

1 lesson1 exercise

Topics & techniques

Keeping it working

Debug by giving the AI the error and the context, and keep changes small enough to stay safe.

1 card →

Workflow

Debug by Giving the AI the Error and the Context

The reliable way to get an AI to fix a bug: instead of saying 'it's broken' or 'that didn't work', you give it the exact error message, what you did to trigger it, what you expected, and the relevant code or files. Precise input turns a guessing game into a targeted fix.

L2 · Context engineering

Test yourself

Can you keep a vibe coded app working?

Three questions on debugging with AI and stopping a small fix from breaking everything around it. No sign-up, instant score, and you can share how you did.

Q1 What does the AI need to fix a bug?
Q2 The AI wants to rewrite a large section to fix a small bug. What's the risk?
Q3 Why does 'that didn't work' make debugging worse?

0 / 3 answered